JP2006506010A - Phase noise reduction device - Google Patents

Abstract

An apparatus for reducing phase noise of a signal (Sin) coming from a quasi-periodic source with a fundamental frequency f0 comprises an active line for transmitting voltage pulses by moving a flux quantum φ0. Includes superconducting circuits. This circuit is defined to have a characteristic frequency fc such that 0.3fc is equal to or lower than the fundamental frequency f0 of the quasi-periodic signal (Sin) applied as an input, and outputs a voltage pulse signal having the fundamental frequency f0. Send out as

Description

  The present invention relates to an apparatus for reducing the phase noise of a signal coming from a quasi-periodic source.

  The invention applies in particular to superconducting logic circuits, in particular logic circuits in RSFQ (High Speed Single Flux Quantum) technology.

  In general, logic systems use at least one clock signal for ordering and synchronization functions. The clock signal is usually generated by an oscillator. These quasi-periodic signals are not completely pure even if the oscillator incorporates a resonant filter. Therefore, when considering the spectral density display of the quasi-periodic signal generated by the oscillator, a noise floor is observed. This is spectral white noise and corresponds to the short-term phase noise of the quasi-periodic signal. In phase-locked circuits normally used in digital systems (computers or other systems), this short-term phase noise reduction is not possible. Its operation has a long-term stabilizing effect to prevent frequency drift.

  In the following, the term “phase noise” shall be understood to mean noise corresponding to noise floor or white noise in the frequency spectrum of the signal. The subject of the present invention is an apparatus for reducing this phase noise. Such a device is particularly useful in the field of high-speed digital electronics. This arrangement allows the reduction of jitter in the clock signal, which is particularly troublesome in digital circuits operating at high and very high frequencies.

  A logic family using superconducting circuits has been developed in high-speed digital electronic systems. This is an RSFQ (High Speed Single Flux Quantum) logic family based on the use of flux quantization and the movement of a single flux quantum φ0. In this approach, logical data processing will eventually manipulate voltage pulses due to the passage of flux quanta in the current loop. One of the basic elements of this logic family based on superconductors is a branched Josephson junction, which allows a single flux quantum to be moved or retained and to the junction. The passage of flux quanta results in a voltage pulse at the end of ∫Vdt = h / 2e = φ0 = 2.07 × 10−15 Weber (h is the Planck constant). Thus, using current technology, the voltage pulse has an amplitude on the order of 2 millivolts over 1 picosecond.

  Each junction is defined by a critical current Ic and a normal resistance Rn depending on its geometry and the technology used. Propagation / movement functionality is provided by appropriate junction bias current control, which allows the current flowing through the junction to be increased or decreased, thus holding the flux quanta in the loop or following through the junction. It is possible to move the magnetic flux quantum to the loop.

RSFQ logic has resulted in many logic circuits, such as analog / digital converters, random access memories, and signal processing processors that compute fast Fourier transforms that can operate at very high frequencies. The upper operating limit of an RSFQ logic element is given by its critical frequency, which depends on the logic element geometry and the technology used (three layers, planar, etc.). This characteristic frequency is given by the following equation: That is,
fc = IcRn / φ0
Here, Ic is a critical current of the junction, Rn is a normal resistance, φ0 is a flux quantum, and is equal to 2.07 × 10 −5 Weber.

  A useful discussion of application examples in RSFQ logic is Konstantin K. Document “Progress and prospects of superconducting electronics” by Konstantin K. Likharev (Superconducting Science Technology (Superconducting Science Technology) (3), Superconducting Science Technology (Superconducting Science 37) It is.

  Another active element in RSFQ logic is the Josephson transmission line. A Josephson transmission line is a line including Josephson junctions branched in parallel, and these junctions are coupled therebetween by a superconducting inductor. Such a line allows the propagation of single flux quanta and thus serves as a transport medium for logical data.

  A very short voltage pulse of the order of 2 millivolts over 1 picosecond applied as the input of such a line propagates along this line through the permanent current loop by propagation of the flux quantum φ0, also called fluxon. . This voltage pulse is recovered at the output. These Josephson transmission lines enable logic pulse transmission without distortion.

  When two pulses are applied continuously as inputs, two fluxons are generated on the line and propagate along this line. These two fluxons are separated on the line by a distance representing the time interval separating the two pulses applied as inputs. However, due to the repulsive interaction between the fluxons that are generated, if the distance d between the two fluxons is so short that this repulsive interaction is significant, spatial redistribution occurs on the line, Is revealed at the output by the time interval separating the two pulses, which is different from the time interval observed at the input of the line. In other words, on the track, one pulse was accelerated and the other was decelerated. This effect is achieved by V.V. K. "Fluxon interaction in an overdamped Josephson transmission line" (Applied Physics Letters, Applied Ps. L, 66) Clearly described in a document entitled June 12, 1995), which includes 200 branched Josephson junctions coupled in parallel by a superconducting inductor and has a characteristic frequency fc of 104 GHz It contains a number of examples of this effect that have been experimentally observed on the Josephson transmission line. Two voltage pulses corresponding to fc-1 and separated by 9.6 ps (picoseconds) are applied as inputs to this line. The time interval between two fluxons propagating along this line increases. At the output, two voltage pulses separated by 27 ps are obtained. Because of the repulsion between the fluxons, one pulse was decelerated and the other was accelerated, resulting in an increase in the time interval separating the two pulses. This correction phenomenon is actually observed only for the junction saturation time for which a value is determined at 3fc-1, i.e., the inter-Fluxon distance corresponding to a time interval less than about 28.8 ps in the example. If the distance between the fluxons is too large, the force is not strong enough. Therefore, the generated fluxon needs to be sufficiently close so that the force is strong. In the example, if two pulses 30 picoseconds apart are injected into the line, this time interval at the output of the line is unchanged.

  Thus, the bit sequence representing the logical data can be modified in the Josephson transmission line due to the repulsive interaction effect between the fluxons, which is equivalent to the loss of logical information. In logic systems, this loss of information can lead to significant effects, ie loss of raw information, desynchronization (phase comparator), and the like. In order to avoid this interaction problem, the author of the document seems that the time interval between the two fluxons generated on the track is 3fc-1 or higher, ie 28.8 ps (saturated) in this example. It is recommended to design the track. Appropriate design is obtained in the circuit definition, especially by changing the critical current, normal resistance and inductance. Thus, the interaction effect can be reduced in operation by changing the Josephson junction bias current.

  Regarding the filtering of white noise in signals coming from quasi-periodic sources, in the present invention, this repulsive interaction effect between fluxons serves to derive advantageous technical effects therefrom. The basic concept of the present invention uses this effect on a series of pulses from a clock signal coming from any quasi-periodic source of fundamental frequency f0 to reduce the white noise level of this signal relative to the fundamental level. Is to do so. Taking the case of a type of clock signal consisting of voltage pulses, the white noise level is due to the temporal dispersion of the signal pulses and thus the dispersion of the spatial distance between the fluxons generated in the superconducting transmission line. Because.

  The interaction effect over the entire length of the track means the following: That is, it is observed that the redistribution of fluxons within a limited distance of the line corresponds to the average value of the distance between fluxons by a large number of random behaviors centered on the smooth value. A direct effect of this spatial redistribution of Flaxson is the temporal redistribution of pulses at the output.

  The white noise of a quasi-periodic signal is manifested in the signal by the temporal dispersion of the pulses and in the superconducting transmission line by the dispersion of the spatial distance between two consecutive fluxons.

  Due to the periodic nature of the signal at the input, the fluxon is configured as a periodic grating in the line. This is a one-dimensional periodic grating along the propagation direction of magnetic flux quantum in the Josephson transmission line. After a certain number of pulses, this grid redistribution occurs corresponding to the transient delay, and the smooth inter-Faxon distance is approximately average.

  Thus, the repulsive phenomenon between fluxons combined with a large number of statistics leads to an even redistribution of fluxons within the lattice, thereby resulting in a white noise level in the quasi-periodic signal at the output of the line. A reduction is brought about.

  In general, according to the present invention, particles having a repulsive interaction between themselves are generated for an interparticle distance shorter than a system saturation value (characteristic frequency) such as an electron (quanttronic circuit), a magnetic flux quantum, or a vortex. Taking any physical system that can do, phase noise can be reduced by reconstructing the particle lattice in that physical system.

  The invention therefore relates to an apparatus for reducing the phase noise of a signal coming from a quasi-periodic source of fundamental frequency f0. According to the present invention, the device includes a physical system for transmitting pulses by moving particles, but this system has an operating frequency range of the device with a lower limit that depends on the characteristic frequency. Defined as having a characteristic frequency fc that defines, as a method, for a quasi-periodic signal applied as input, the particles have a repulsive interaction with each other and the system outputs a pulse at the fundamental frequency f0 To be sent out.

  The invention also relates to a device for reducing the phase noise of a signal coming from a quasi-periodic source of fundamental frequency f0. In accordance with the present invention, the device includes a superconducting circuit with an active line for transmitting voltage pulses by the movement of the flux quantum φ0, and the circuit has 0.3fc ≦ f0 (where f0 is defined to have a characteristic frequency fc which is a fundamental frequency of a quasi-periodic signal (Sin) applied as an input, and a voltage pulse signal having a fundamental frequency f0 is transmitted as an output.

  Phase noise reduction can be improved by defining a superconducting circuit consisting of active voltage pulse transmission lines, in which case the flux quanta generated in the circuit by the effect of applying a quasi-periodic signal is It is constructed along a dimensional periodic lattice. Thus, interactions between flux quanta occur between the nearest flux quanta along the two dimensions of the lattice.

  The invention applies not only to flux quanta generated in Josephson transmission lines, but more generally to any superconducting circuit based on active voltage pulse transmission lines. In particular, the present invention also provides a vortex flux transmission line, i.e. a transmission line, comprising a long Josephson junction, a Josephson vortex flux flow, a slot line or a microbridge line, or an Abrikosov vortex flux flow. Applies to

  The phase reduction device can also be advantageously used in a frequency multiplier circuit.

  Other advantages and features of the present invention will become more clearly apparent when reading the following description given as a non-limiting representation of the invention with reference to the accompanying drawings.

  FIG. 1 shows the spectral density A (Sin) of a signal Sin coming from a quasi-periodic source and applied as a clock signal in a logic system. In the present invention, the goal is to reduce the phase noise / signal ratio N2 / N1, which is about −115 to −120 dBc for signals coming from a conventional quasi-periodic source (oscillator), to at least 1/10. . Such a reduction is particularly advantageous in the field of electronics operating at very high frequencies, and in particular in systems based on high Tc (high critical temperature) superconducting RSFQ logic circuits with low thermal noise. Thus, the benefits of signals with significantly reduced short-term noise are fully utilized.

  FIG. 2 shows a first embodiment of a phase noise reduction device according to the invention comprising a superconducting circuit based on a voltage pulse transmission line, the signal Sin to be processed being applied to the input of this circuit, Then, the circuit sends out a signal Sout with reduced phase noise as an output.

  In this example, the transmission line includes a plurality of Josephson junctions JJ1, JJ2,. . . This is a Josephson transmission line including JJ200. Josephson junctions are branched and mounted in parallel, and superconducting inductors Ls1, Ls2, Ls3,. . . They are coupled to each other via Ls200. A superconducting inductor Ls0 is also provided at the input portion between the input signal electrode A and the first Josephson junction JJ1. The input signal is applied between the two input signal electrodes A and M at the end of the line. The output signal Sout is obtained at the output part of the line between the two output signal electrodes B and M ′. Electrodes M and M 'are line ground electrodes. The junction is biased with a current Ic less than the critical current of the junction so that a permanent current loop Bc is established for each cell closed by the junction.

  When a pulse is applied to the input portion of such a line, the junction current increases beyond the critical current. Josephson effect occurs. That is, the flux quanta pass through the current loop and a corresponding voltage pulse appears at the end of the junction. Thus, the voltage pulse propagates along the line without being distorted.

  When a clock signal pulse train is applied, the corresponding train is recovered at the output. According to the present invention, the characteristics of the line are selected so as to obtain a predetermined characteristic frequency fc. This characteristic frequency fc defines the operating frequency range of the device with a lower limit that depends on this characteristic frequency. The quasi-periodic signal applied to the input has its fundamental frequency within the operating range defined in this way, but with this quasi-periodic signal, an effective repulsive interaction is obtained, thereby It is possible to reduce the background of white noise of this signal.

  In particular, the characteristics of the line are selected so as to obtain a characteristic frequency fc that satisfies 0.3fc ≦ f0. Here, 0.3 fc is the lower limit of the operating range of this device.

  Therefore, on average, the distance between fluxons is less than the saturation value of the line. The repulsive interaction phenomenon between flux quanta (fluxons) results in a spatial redistribution of flux quanta (fluxons) along the line around the mean inter-fluxon value, which is the time interval between two pulses. This is done by smoothing around the average value corresponding to the average value. In the output section, the signal has a considerably reduced standard deviation of the time interval between pulses. In this way, short-term noise or phase noise of the input signal is reduced.

  The characteristic of the Josephson transmission line is mainly the inductance, which depends on the length of the line and the technology, in particular the mutual inductance Lm, and the characteristics of the junction, ie the critical current Ic and the normal resistance Rn. In order not to overly complicate the drawing of FIG. 2, these well-known properties of the Josephson junction are shown only for the first junction JJ1.

  FIG. 3 shows a practical embodiment of the phase reduction device according to the present invention having a Josephson transmission line type superconducting circuit. In this embodiment, a thin film of a high Tc superconductor on a bicrystal substrate is shown. Multiple Josephson junctions in the based planar technology are included.

  Two substrates 1 and 2, which are typically SrTiO 3 substrates or MgO or YSZ substrates, are bonded together, but the crystal axes of these substrates have an angular difference in the surface plane. A superconducting film 3, typically a YBa 2 Cu 3 On (where 6 ≦ n ≦ 7) shaped film, is deposited (by epitaxy) on the surface plane of the bicrystal across the bond line of the bicrystal substrate. As a result, the grain boundary 4 grows just under the superconducting film along the bond and becomes equal to the electrical barrier. The film is then etched into a ladder pattern, and each step of the ladder corresponds to a Josephson junction.

  In this example, the step width w is about 5 microns, the step length l is about 20 microns, and the spacing h between the two steps is on the same order (20 microns). The membrane itself has a width of a few microns for a thickness of a few tenths of a micron (eg, 0.3 μm). The thickness of the substrate is several hundred microns, typically 300-1000 μm.

  A current source (not shown) delivers a bias current typically on the order of 100 microamps to each of the Josephson junctions for the technique taken as an example. In this example, this bias current is applied between two current bias electrodes C and C ′ formed in the portion 3 ′ of the superconducting film 3. This part is arranged on both sides of the ladder forming a series of junctions and the pairs b1, b'1. . . The current supply branch provided at b100, b'100 is shaped (by etching) to distribute this current just along the line. In this example, the current supply branch b1 and its complementary, ground line side branch b'1 apply a bias current to the two junctions JJ1 and JJ2 located on both sides of these branches. For a line containing 200 Josephson junctions, the current source is designed to deliver a bias current on the order of tens of milliamps (eg, 20 mA) distributed along the line.

  Input and output signal electrodes A, M, B, M ′, typically made of copper or gold, are formed at each end of the film and on both sides of the grain boundaries 4.

  For example, including 200 junctions, a length of about 2 millimeters, a seaside junction current Ic of 125 microamps, and a characteristic frequency fc (where fc = IcRn / φ0 = 125 × 10−6 × 2 / 2.07 × A Josephson transmission line with a normal resistance Rn of 2 ohms defining 10-15 Weber = 116 gigahertz is required for a high critical temperature of 30 Kelvin or less and a 100 microampere bias current (<Ic) for each junction. ) And a niobium superconducting film (0.1 μm thin film). When a clock signal with a fundamental frequency f0 (≧ fc / 3) of about 50-100 gigahertz and having a very offset (short term noise) pulse over time is applied to this line as input, white noise / It is possible to supply as output the Sout of the signal whose signal ratio has been reduced by a factor of ten, i.e. about -130 to -140 dBc (rather than -115 to -120 dBc at the input).

  FIG. 4a schematically shows the lattice structure of a fluxon generated in such a line under the influence of a voltage pulse signal Sin applied as input.

  When the line is displayed as channel 5, the voltage pulse of signal Sin is injected into one end of this channel at clock frequency f0. Flaxons flx1, flx2,. . . flxm is generated in channel 5 and is spatially constructed along a one-dimensional lattice corresponding to the direction of fluxon propagation in the line.

  Since transmission lines, ie lines containing a large number of junctions (as opposed to superconducting logic circuits of the type including only a small number of junctions, such as shift registers) are used, two input signals are used. By smoothing the inter-Fluxon distance around an average value d0 that matches the average value of the time interval between pulses, a spatial redistribution effect is produced. In other words, the standard deviation of the time interval value between pulses in the output signal is reduced. More precisely, as shown in FIG. 4b, the phase noise of the signal Sin applied as input is revealed by a distributed temporal distribution in this signal. As schematically shown in FIG. 4b, the fluxon generated by this signal is also spatially distributed in this line. The line characteristic (fc) is selected so that the distance between fluxons generated by the input signal Sin is on average smaller than the saturation value of the line, so that there is a repulsive interaction between the nearest adjacent fluxons. There is. In the figure, these repulsive forces are indicated by arrows. In the example shown in this figure, the saturation value is assumed to correspond to a time difference of 22 picoseconds. Thus, whenever the fluxon distance corresponds to a time difference less than this value, the mutual repulsion produces its (flx1-flx2, flx2-flx3, flx4-flx5) effect. If this distance is larger, there is no (flx3-flx4) effect. In practice, after a transient phase corresponding to about 20 pulses, the fluxon is spatially redistributed in the line around the smoothed value of the inter-flaxon distance. In the example schematically shown in FIG. 4c, this smoothed value corresponds to a time difference between two pulses of the output signal Sout of 20 picoseconds.

  In this way, the output signal is distributed more evenly than the signal level at the fundamental frequency f0, which corresponds to a reduction in the phase noise level. In practice, using a transmission line as shown in FIG. 3, a 1/10 reduction in the N2 / N1 ratio (FIG. 1) can be observed.

  Spatial separation, and thus interaction, depends on the ratio of fluxon propagation speed to signal frequency. The fluxon speed can be changed by modifying the bias current. Thus, the bias current may be adjusted according to the frequency of the input signal if such a need is necessary.

  Figures 5a and 5b show an alternative embodiment of a phase reduction device based on a superconducting Josephson transmission line circuit. In the present embodiment, the superconducting circuit includes two Josephson transmission lines. Further, the substrate 1 and the substrate 1 ′ are bonded to both sides of the substrate 2 to form a tricrystal substrate. A superconducting film is deposited in zones 3a and 3b, one of which is on each bond line to grow the respective grain boundaries 4a, 4b. In these figures, the current supply branches distributed along the track are wires, typically copper wires, that coincide with the contact pads 6 provided on the membrane.

  In such a configuration, it is possible to improve the effectiveness of spatial redistribution in the line by adding another dimension to the interaction phenomenon between fluxons. The same interaction phenomenon is observed by placing the membranes in zones 3a and 3b separated by a gap so that the distance between the fluxon in one membrane and the fluxon in the other membrane is less than the saturation value. Is done. In other words, for a superconducting circuit based on two Josephson transmission lines, the fluxon generated in the circuit is constructed along a two-dimensional periodic grating. Typically, a gap of a few microns must be provided for the numerical examples of line characteristics and frequency (f0) values described above.

  To make this effect effective, it is necessary to choose a stable (staggered) configuration of a fluxon two-dimensional periodic grating, typically for a superconducting circuit based on a triangle base. Otherwise, repulsion may have a random effect in the line propagation direction x or in the opposite direction. This is therefore an unstable situation. Referring to FIG. 5a, where the two membranes forming the Josephson transmission line are perfectly aligned along x and y, the desired grating applies the signal applied as input to the second line, π Only by phase shifting. As shown in FIG. 5b, a two-dimensional triangular-based periodic grating is obtained. Next, the fluxon flx of the line has four fluxons, that is, two fluxons flx1 and flx2 on either side of this fluxon flx on the same line, and a bisector 7 of this line passing through the fluxon flx on the other line. Are interacted by two fluxons flx3 and flx4 located on both sides of the.

  As shown in FIGS. 6a and 6b, the π phase shift may be applied in various ways.

  In FIG. 6a, a π phase shift is added to the input signal Sin. Furthermore, the signal coming from the quasi-periodic source 100 is preferably applied to the circuit 101 so that it is split in two as output. An example of this splitter circuit 101 resulting in RSFQ logic is shown in detail in the figure as a practical example. It sends two signals in phase as outputs.

  In FIG. 6b, a π phase shift is added to the output signal Sout, l of the first line, and this signal is injected into the second line. In this case, the fluxon at the beginning of the first line benefits from the spatial redistribution already obtained at the output of this first line-this is a positive feedback effect. Furthermore, an interconnect line 102 is provided for supplying the output signal from the first line as an input for the phase shifter of the second line. This line is typically made of coplanar, strip or microstrip type technology, using materials that are compatible with the Josephson transmission line used, or even a Josephson transmission line. Good.

  The two Josephson transmission lines may not be accurately aligned on the substrate, and the interconnect line 102 introduces a delay such that the output signals Sout, l and Sout, 2 are not completely π phase shifted. there is a possibility. In this case, the interaction between the lines may not be optimal. Advantageously, the bias current Ib of one or more junctions can be advantageously modified locally in order to adapt the fluxon propagation velocity locally. This correction is easily applied by the current supply branch (FIG. 3) or the current supply wire (FIG. 5a) just because the current is distributed along the line. In this way, provision is made to make the bias current Ib of the junction preferably variable, which can be adjusted for each junction or group of junctions.

  It is also possible to provide more than two transmission lines in parallel on the surface plane of the substrate. FIG. 6c shows an example of a circuit including three Josephson transmission lines. For the positive line-to-line interaction effect, the displacement of the fluxon along the line propagation direction x is preferred, but in order to obtain the effect, the central line Li1 that receives the input signal Sin as input, and (shown in FIG. 6a) The two lines Li2 and Li3 on either side of the central line receiving the π phase shifted signal as input, which may be the input signal Sin or the first line output signal Sout, 1 (FIG. 6b), Is provided. Again, the fluxon is constructed along a two-dimensional periodic grating, but the number of lines in this grating is increased. Thus, the fluxon of the central line Li1 is the interaction from its own line and the interaction by the other two lines, ie for each fluxon, two per line, six nearest neighbors. Receive up to 6 interactions.

  By increasing the number of lines in parallel, the number of interactions is increased. In the three-line example (FIG. 6c), the central line Li1 benefits from the interaction by the two lines Li2 and Li3 located on both sides of the center line Li1, but the lines Li2 and Li3, respectively, only from the interaction by the line Li1. Profit.

  Selecting many lines depends on the design of the device that is acceptable in the application. In the case of parallel n-lines, each line is shorter, i.e. fewer junctions, due to the retroactive effect of redistribution combined with the additional dimension of interaction in the two-dimensional lattice thus formed. It should be noted that it may be. The design is evaluated in such a way that a large number of statistics can be applied to produce the desired effect of smoothing the fluxon distance.

  In general, in the case of parallel n-lines, the signals are alternating, ie the input signal to one line, then the input signal phase-shifted to the next line (by phase shifter circuit—FIG. 6a) and so on. Applied. For example, an even-order line receives an input signal (Sin), and an odd-order line receives an input signal that is phase-shifted. The output signal of the device is sent as output from one of the lines.

  FIG. 7 shows an example of a phase noise reduction device used in the frequency doubler circuit. In this example, the circuit includes two lines in parallel, the first line receives the input signal Sin, and the other line receives the phase-shifted input signal. The first line sends signals Sout, 1 as output, and the other line sends signals Sout, 2 as output.

  The two lines are arranged in such a way that the line fluxons interact with each other to reduce short-term phase noise. The two output signals Sout, 1 and Sout, 2 obtained as outputs in this way are applied as inputs to an RSFQ (coupler) logic circuit, which has low phase noise and the frequency of the input signal Sin. A signal S (2f0) having a frequency twice as high is sent as an output.

  Thus, the phase noise reduction device according to the invention advantageously uses this type of circuit cascading in frequency doubler circuits, more generally in frequency multiplier circuits, yet very low It is possible to maintain a background of phase noise.

  FIG. 8a shows another example of one embodiment of a Josephson transmission line, which can be used in all alternative embodiments of the phase reduction device according to the invention just described. FIG. 8b may be used in a single line or a structure consisting of multiple lines stacked vertically. In these two FIGS. 8a and 8b, the lines are made with lamp edge bonding technology, which is a SNS (representing superconductor / normal or insulating material / -superconductor) multilayer technology. The normal or insulating material is, for example, PrBaCuO, which is a non-superconductor, but this material has a similar structure to YBaCuO that can coexist with the lattice cell properties of the superconductor. The comb shape on the substrate includes a first superconducting film 9 (thin film) deposited on a heterostructure (8) of normal or insulating material deposited on a superconducting base electrode. The structure is shown in gray in the figure. The comb teeth have a ramp shape that decreases toward the substrate. A thin layer of insulator in the shape of a comb and a second superconducting film 10 are deposited on the substrate, the ends of the teeth of the comb being on the ends of the teeth of the superconducting film of the first comb To do. Therefore, the junctions JJ1, JJ2,. . . , Etc. are formed in a plane between the two superconductor films 9 and 10 where the normal or insulating material layer 8 is thinned.

  FIG. 8 b is a variation of FIG. 8 a where the second superconductor film 10 is “folded” on the first film 9, which allows a considerable amount of area savings.

  FIG. 9a shows another embodiment of a phase noise reduction device consisting of a superconducting circuit based on a voltage pulse transmission line. In this embodiment, the transmission line is produced by a long Josephson junction. Such a junction is typically obtained with SIS three-layer technology, preferably based on low Tc superconductors. That is, the thin film 20 of normal (or insulating) material (eg, Al 2 O 3) forms a barrier between the two layers 21 and 22 of superconductor (eg, niobium). A bias current i smaller than the critical current Ic of the long Josephson junction is applied between the two superconductor layers 21 and 22. When a pulse is applied to the input section of the line, a vortex (Josephson vortex) magnetic flux is generated in the normal material layer, which is output under the influence of the bias (direct current) current (Lorentz force) of the line. Propagate to. The flux quantum associated with each vortex is equal to φ0. The same repulsive interaction effect applies to the vortex flux generated under the influence of the clock signal Sin, but these vortex fluxes are composed of the line as a one-dimensional periodic lattice, along the propagation direction x of the line. Propagate.

  In an exemplary embodiment, the length of such a line is about 100 nanometers.

  Some of these lines may be stacked vertically as shown in FIG. 9b and placed in parallel to achieve the same advantageous effect seen above, but this is feasible. Even if it is, it is more difficult to handle.

  The current is preferably distributed along the line as shown in FIG. 9b. The level of the bias current may be adjusted according to the frequency of the input signal.

  Another embodiment of a phase noise reduction device according to the present invention is shown in FIGS. 10a and 10b, but this embodiment is based on a type II superconductor circuit based on an active Abrycosov vortex flux flow transmission line. Equivalent to. The Abrikosov vortex magnetic flux principle is simply as follows. That is, in the presence of an increasing magnetic field, the superconductor switches to the normal conductor / superconductor hybrid state. A current is generated on the surface of the superconductor that tends to shield the magnetic field. The magnetic flux entering the superconductor is in the form of field lines grouped together on the surface of a disk with a radius of tens of angstroms. The magnetic flux contained in this small zone surrounded by the magnetic field shield current circulating around the magnetic flux is equal to the flux quantum φ0. As shown in FIG. 11, these vortex magnetic fluxes are formed on the surface as a triangular base periodic grating. By injecting a direct current directed in the proper direction, this vortex flux grid propagates in translation along the direction perpendicular to the current (Lorentz force).

  One advantage of such a transmission line is that the vortex flux is “naturally” configured as a triangle-based two-dimensional periodic grating.

  By appropriately applying a bias current to the device, application of an electromagnetic signal as an input creates a vortex flux grid that travels along the line Lv (FIG. 11) along this grid structure. At the output, the receiving device (any matched load) receives the associated voltage pulse.

  Furthermore, in a superconducting material used, such as NdBa2Cu3O7, this configuration makes sense if a pair of planes are arranged in parallel. That is, the line Lv corresponds to two planes. According to the present invention, the active superconductor circuit includes a second type superconductor film (thin layer) 13 such as YBa2Cu3O7 or NdBa2Cu3O7 deposited (by epitaxy) on a substrate 12 such as a SrTiO3 substrate (FIG. 10a, 10b). A slot 14 is created across the width of the membrane leaving only the superconducting membrane microbridge 15 between the two portions 13a and 13b of the membrane on either side of the slot. The height of this microbridge is equal to or less than the thickness of the membrane. In this example, the microbridge has a height e of approximately 0.1 microns, a microbridge length L along the slot direction of less than 100 microns, and a width W that is also the width of the slot is greater than 100 microns. .

  Two bias electrodes 16 and 17 are provided at each end of the membrane to apply a DC current i of approximately a few milliamps. Two input signal electrodes 18 and 19 are provided at one end of the slot at each membrane portion 13a, 13b on either side of the slot. This is because an AC input signal Sin is applied so that the input signal periodically imposes a local magnetic field Be larger than the critical magnetic field at the input portion of the microbridge, and a vortex v is generated at the period of this signal. It is to do. The input signal may be a voltage pulse signal. It is also possible to apply a sinusoidal AC signal. In practice, the clock signal source (not shown) is impedance matched to the impedance (several tens of ohms) of the microbridge.

  In order to receive as input the voltage pulses corresponding to the in-line transmission of the vortex (FIG. 11), the two output signal electrodes 20 and 21 are connected to the other side of the slot at each part 13a, 13b of the membrane on both sides of the slot. Is provided at the end.

  In practice, each voltage pulse (or each positive peak voltage of the AC signal) passes as a microbridge input through a local magnetic field Be that exceeds the critical field of the superconducting film, causing the vortex group to nucleate. To do. A direct current i (Lorentz force) applied at right angles along the appropriate direction moves the vortex.

  A vortex is generated by adjusting the magnetic field with a clock signal applied as an input. By applying an appropriate bias to the circuit, the vortex is propagated along the desired direction towards the output Sout of the device.

  To further promote the displacement of the vortex in the desired direction, place the device in a low direct current magnetic field B in an appropriate orientation, for example at about 20 millitesla, for example by placing a pair of Helmholtz coils on both sides of the circuit, It is possible to direct the vortex in the same direction.

  Such a superconducting circuit may be advantageously used in the frequency doubler stage shown above and another similar circuit associated with the phase shifter circuit in the frequency multiplier.

  As described above, in the present embodiment, the transmission line includes the hybrid type second-type superconductor film deposited on the crystal substrate. The membrane is biased at its ends and includes a slot in the width direction, except at the microbridge, and this slot separates the membrane into two parts. A quasi-periodic signal is applied to one end of the slot between the two portions of the membrane and an output signal is obtained at the other end of the slot between the two portions of the membrane.

  Advantageously, such a superconductor device is immersed in a DC magnetic field perpendicular to the surface plane of the slot.

  Thus, the invention just described describes the periodic structure of the generated flux quanta (fluxons, vortex) lattices and the repulsive interaction properties of these flux quanta (these can be compared to magnetic dipoles). ) To reduce the phase noise of the signal coming from the quasi-periodic source. This device according to the invention is advantageously used to deliver multi-frequency signals without any deterioration of phase noise.

  The present invention is particularly applicable to high frequency and ultra high frequency fields in high speed electronic systems. Such a device may be used in particular in RSFQ logic circuits.

As already explained, the spectral density A (Sin) of the signal coming from the quasi-periodic source is shown. 1 shows a circuit diagram of a phase reduction device according to the invention based on a Josephson transmission line including a plurality of Josephson junctions. FIG. A first example of such a line embodiment in a bicrystal multi-junction technique is shown. In the Josephson transmission line, the periodic grating of the fluxon generated by the pulse clock signal is schematically shown, and the temporal redistribution phenomenon of the voltage pulse in such a line is shown. FIG. 5 shows another example of an embodiment of a phase reduction device that includes two Josephson transmission lines arranged in parallel in the same surface plane, and is a corresponding illustration of a fluxon periodic grating. In order to improve the effectiveness of the correction, two alternative ways of using two Josephson transmission lines in parallel in the phase reduction device are schematically shown, with n = 3 in parallel, along with a corresponding illustration of the fluxon periodic grating. It is an alternative to the previous figure with a Josephson transmission line. An example in which a phase noise reduction device is used in a frequency doubling circuit is shown. Another example of a phase reduction device based on a Josephson transmission line made with lamp edge bonding technology is shown. 2 shows two embodiments of a phase noise reduction device based on a long Josephson junction transmission line. 1 shows a phase noise reduction device based on vortex flux, slot or microbridge line. It is an illustration of a periodic lattice of vortex yarn generated in such a line.

Claims (19)

A device for reducing the phase noise of a signal (Sin) coming from a quasi-periodic source of fundamental frequency f0,
The apparatus includes a physical system for transmitting pulses by moving particles, the physical system defining a characteristic frequency fc that defines an operating frequency range of the apparatus with a lower limit that depends on the characteristic frequency. The method is such that the particles have a mutual repulsive interaction with respect to the quasi-periodic signal (Sin) applied as input, and the system delivers a pulse at the fundamental frequency f0 as output. A device characterized by that. An apparatus according to claim 1 for reducing phase noise of a signal (Sin) coming from a quasi-periodic source of fundamental frequency f0,
The device comprises a superconducting circuit with an active line for transmitting voltage pulses by moving the flux quantum φ0, the circuit comprising the quasi-periodic signal (Sin) with 0.3 fc applied as input And a voltage pulse signal having a fundamental frequency f0 that is defined to have a characteristic frequency fc that is equal to or lower than the fundamental frequency f0. At least two superconducting circuits, i.e., a circuit for pi-phase-shifting the input or output of one of said circuits, and a combiner circuit for providing a frequency doubling stage in the frequency multiplication circuit. The phase noise reduction device according to claim 1 or 2. The phase noise reduction device according to claim 2 or 3, wherein the superconducting circuit includes a Josephson transmission line that is geometrically defined by the characteristic frequency. The phase noise reduction device according to claim 4, wherein the Josephson transmission line is a long Josephson junction.   The phase noise reduction device according to claim 4, wherein the transmission line includes a plurality of Josephson junctions branched in parallel. The phase noise reduction device according to claim 6, wherein each Josephson transmission line includes a line having a bicrystal junction.   The phase noise reduction device according to claim 6, wherein each Josephson transmission line is of a type including a line having a ramp edge junction. The phase noise reduction device according to any one of claims 5 to 8, wherein the superconducting circuit includes a number of Josephson transmission lines arranged in parallel. The phase noise reduction device according to claim 9, further comprising a π phase shift circuit at an input unit of at least one transmission line, and applying a phase-shifted signal to the line. The phase noise reduction device according to claim 10, wherein the phase shift circuit receives the input signal (Sin) of the device as an input. The phase noise reduction device according to claim 10, wherein the phase shift circuit receives an output signal from a line as an input. The superconducting circuit includes n Josephson transmission lines in rows 1 to n on the same surface plane of the substrate, and n is an integer of 2 or more;
One of the input signal and the phase-shifted input signal is applied to an even-numbered line, the other signal is applied to an odd-numbered line, and the output signal is The phase noise reduction device according to claim 11, wherein the phase noise reduction device is sent as one output. 14. A current biasing means including a plurality of branches for supplying current, the current being distributed along each Josephson transmission line. The phase noise reduction device according to 1. The phase noise reduction device according to any one of claims 1 to 14, further comprising means for adjusting an intensity of the bias current according to a frequency of the input signal. The phase noise reduction device according to any one of claims 2 to 4, wherein the superconducting circuit includes a vortex magnetic flux flow voltage pulse transmission line. The transmission line includes a hybrid type II superconducting film deposited on a crystal substrate, and the film is biased at an end thereof and is slotted in the width direction except for a microbridge. Wherein the slot separates the membrane into two parts, and the quasi-periodic signal is applied to one end of the slot between the two parts of the membrane, and the output signal The phase noise reduction device according to claim 16, characterized in that is obtained at the other end of the slot between the two parts of the membrane. The phase noise reduction device according to claim 16 or 17, wherein the superconducting device is immersed in a DC magnetic field oriented in a direction perpendicular to a surface plane of the slot. The phase noise reduction device according to any one of claims 1 to 18, wherein the one or more superconducting circuits use a high critical temperature superconductor.

Images